报错 No available block found in 60 second.

[948024326]

System Info / 系統信息

4*32g v100

Running Xinference with Docker? / 是否使用 Docker 运行 Xinfernece？

• docker / docker
• pip install / 通过 pip install 安装
• installation from source / 从源码安装

Version info / 版本信息

0.16.2

The command used to start Xinference / 用以启动 xinference 的命令

nohup启动

Reproduction / 复现过程

报错过程
vllm部署模型 4卡 之前一直好好的 然后调用会话后出现 第一张卡直接的gpu进程没了 然后其他三张卡模型还挂着 但是调用会话就一直没响应 只能进行重启服务再挂载模型

报错信息如下

ERROR 11-27 11:41:33 async_llm_engine.py:64] Engine background task failed
ERROR 11-27 11:41:33 async_llm_engine.py:64] Traceback (most recent call last):
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 54, in _log_task_completion
ERROR 11-27 11:41:33 async_llm_engine.py:64] return_value = task.result()
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 851, in run_engine_loop
ERROR 11-27 11:41:33 async_llm_engine.py:64] result = task.result()
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 774, in engine_step
ERROR 11-27 11:41:33 async_llm_engine.py:64] request_outputs = await self.engine.step_async(virtual_engine)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 346, in step_async
ERROR 11-27 11:41:33 async_llm_engine.py:64] outputs = await self.model_executor.execute_model_async(
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/distributed_gpu_executor.py", line 181, in execute_model_async
ERROR 11-27 11:41:33 async_llm_engine.py:64] return await self._driver_execute_model_async(execute_model_req)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/multiproc_gpu_executor.py", line 224, in _driver_execute_model_async
ERROR 11-27 11:41:33 async_llm_engine.py:64] return await self.driver_exec_model(execute_model_req)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/concurrent/futures/thread.py", line 52, in run
ERROR 11-27 11:41:33 async_llm_engine.py:64] result = self.fn(*self.args, **self.kwargs)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 303, in execute_model
ERROR 11-27 11:41:33 async_llm_engine.py:64] inputs = self.prepare_input(execute_model_req)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 291, in prepare_input
ERROR 11-27 11:41:33 async_llm_engine.py:64] return self._get_driver_input_and_broadcast(execute_model_req)
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 253, in _get_driver_input_and_broadcast
ERROR 11-27 11:41:33 async_llm_engine.py:64] self.model_runner.prepare_model_input(
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1586, in prepare_model_input
ERROR 11-27 11:41:33 async_llm_engine.py:64] model_input = self._prepare_model_input_tensors(
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1196, in _prepare_model_input_tensors
ERROR 11-27 11:41:33 async_llm_engine.py:64] return builder.build() # type: ignore
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 867, in build
ERROR 11-27 11:41:33 async_llm_engine.py:64] attn_metadata = self.attn_metadata_builder.build(
ERROR 11-27 11:41:33 async_llm_engine.py:64] File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/attention/backends/utils.py", line 215, in build
ERROR 11-27 11:41:33 async_llm_engine.py:64] input_block_tables[i, :len(block_table)] = block_table
ERROR 11-27 11:41:33 async_llm_engine.py:64] ValueError: could not broadcast input array from shape (516,) into shape (512,)
2024-11-27 11:41:33,903 xinference.core.model 931290 ERROR [request 7b8ea0e6-ac71-11ef-9b19-fa163ea8cbc1] Leave chat, error: could not broadcast input array from shape (516,) into shape (512,), elapsed time: 6 s
Traceback (most recent call last):
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/utils.py", line 78, in wrapped
ret = await func(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 709, in chat
response = await self._call_wrapper_json(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 517, in _call_wrapper_json
return await self._call_wrapper("json", fn, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 122, in _async_wrapper
return await fn(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 526, in _call_wrapper
ret = await fn(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/utils.py", line 30, in _async_wrapper
return await fn(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/core.py", line 706, in async_chat
c = await self.async_generate(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/utils.py", line 30, in _async_wrapper
return await fn(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/core.py", line 597, in async_generate
async for request_output in results_generator:
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 1029, in generate
async for output in await self.add_request(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 112, in generator
raise result
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 54, in _log_task_completion
return_value = task.result()
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 851, in run_engine_loop
result = task.result()
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 774, in engine_step
request_outputs = await self.engine.step_async(virtual_engine)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 346, in step_async
outputs = await self.model_executor.execute_model_async(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/distributed_gpu_executor.py", line 181, in execute_model_async
return await self._driver_execute_model_async(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/multiproc_gpu_executor.py", line 224, in _driver_execute_model_async
return await self.driver_exec_model(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/concurrent/futures/thread.py", line 52, in run
result = self.fn(*self.args, **self.kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 303, in execute_model
inputs = self.prepare_input(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 291, in prepare_input
return self._get_driver_input_and_broadcast(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 253, in _get_driver_input_and_broadcast
self.model_runner.prepare_model_input(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1586, in prepare_model_input
model_input = self._prepare_model_input_tensors(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1196, in _prepare_model_input_tensors
return builder.build() # type: ignore
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 867, in build
attn_metadata = self.attn_metadata_builder.build(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/attention/backends/utils.py", line 215, in build
input_block_tables[i, :len(block_table)] = block_table
ValueError: could not broadcast input array from shape (516,) into shape (512,)
2024-11-27 11:41:33,909 xinference.api.restful_api 929546 ERROR [address=0.0.0.0:36473, pid=931290] could not broadcast input array from shape (516,) into shape (512,)
Traceback (most recent call last):
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/api/restful_api.py", line 1998, in create_chat_completion
data = await model.chat(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/context.py", line 231, in send
return self._process_result_message(result)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/context.py", line 102, in _process_result_message
raise message.as_instanceof_cause()
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/pool.py", line 659, in send
result = await self._run_coro(message.message_id, coro)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/pool.py", line 370, in _run_coro
return await coro
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/api.py", line 384, in on_receive
return await super().on_receive(message) # type: ignore
File "xoscar/core.pyx", line 558, in on_receive
raise ex
File "xoscar/core.pyx", line 520, in xoscar.core._BaseActor.on_receive
async with self._lock:
File "xoscar/core.pyx", line 521, in xoscar.core._BaseActor.on_receive
with debug_async_timeout('actor_lock_timeout',
File "xoscar/core.pyx", line 526, in xoscar.core._BaseActor.on_receive
result = await result
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 98, in wrapped_func
ret = await fn(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/api.py", line 462, in _wrapper
r = await func(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/utils.py", line 78, in wrapped
ret = await func(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 709, in chat
response = await self._call_wrapper_json(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 517, in _call_wrapper_json
return await self._call_wrapper("json", fn, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 122, in _async_wrapper
return await fn(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/core/model.py", line 526, in _call_wrapper
ret = await fn(*args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/utils.py", line 30, in _async_wrapper
return await fn(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/core.py", line 706, in async_chat
c = await self.async_generate(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/utils.py", line 30, in _async_wrapper
return await fn(self, *args, **kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/model/llm/vllm/core.py", line 597, in async_generate
async for request_output in results_generator:
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 1029, in generate
async for output in await self.add_request(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 112, in generator
raise result
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 54, in _log_task_completion
return_value = task.result()
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 851, in run_engine_loop
result = task.result()
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 774, in engine_step
request_outputs = await self.engine.step_async(virtual_engine)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/engine/async_llm_engine.py", line 346, in step_async
outputs = await self.model_executor.execute_model_async(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/distributed_gpu_executor.py", line 181, in execute_model_async
return await self._driver_execute_model_async(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/executor/multiproc_gpu_executor.py", line 224, in _driver_execute_model_async
return await self.driver_exec_model(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/concurrent/futures/thread.py", line 52, in run
result = self.fn(*self.args, **self.kwargs)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 303, in execute_model
inputs = self.prepare_input(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 291, in prepare_input
return self._get_driver_input_and_broadcast(execute_model_req)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/worker_base.py", line 253, in _get_driver_input_and_broadcast
self.model_runner.prepare_model_input(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1586, in prepare_model_input
model_input = self._prepare_model_input_tensors(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 1196, in _prepare_model_input_tensors
return builder.build() # type: ignore
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/worker/model_runner.py", line 867, in build
attn_metadata = self.attn_metadata_builder.build(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/vllm/attention/backends/utils.py", line 215, in build
input_block_tables[i, :len(block_table)] = block_table
ValueError: [address=0.0.0.0:36473, pid=931290] could not broadcast input array from shape (516,) into shape (512,)

2024-11-27 11:41:34,736 xinference.model.llm.vllm.utils 931290 INFO Detecting vLLM is not health, prepare to quit the process
2024-11-27 11:41:34,736 xinference.model.llm.vllm.core 931290 INFO Stopping vLLM engine
INFO 11-27 11:41:34 multiproc_worker_utils.py:133] Terminating local vLLM worker processes
2024-11-27 11:41:35,202 xinference.api.restful_api 929546 ERROR Remote server 0.0.0.0:36473 closed
Traceback (most recent call last):
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xinference/api/restful_api.py", line 1998, in create_chat_completion
data = await model.chat(
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/context.py", line 230, in send
result = await self._wait(future, actor_ref.address, send_message) # type: ignore
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/context.py", line 115, in _wait
return await future
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/context.py", line 106, in _wait
await asyncio.shield(future)
File "/ai/anaconda3/envs/xinference0161/lib/python3.10/site-packages/xoscar/backends/core.py", line 84, in _listen
raise ServerClosed(
xoscar.errors.ServerClosed: Remote server 0.0.0.0:36473 closed
2024-11-27 11:41:36,476 xinference.core.worker 929804 WARNING Process 0.0.0.0:36473 is down.
(VllmWorkerProcess pid=931684) WARNING 11-27 11:42:27 shm_broadcast.py:396] No available block found in 60 second.
(VllmWorkerProcess pid=931682) WARNING 11-27 11:42:27 shm_broadcast.py:396] No available block found in 60 second.
(VllmWorkerProcess pid=931683) WARNING 11-27 11:42:27 shm_broadcast.py:396] No available block found in 60 second.

Expected behavior / 期待表现

修复bug

[948024326]

@qinxuye

[qinxuye]

vLLM 似乎跑挂了，他有自动重启吗？

[948024326]

vLLM 似乎跑挂了，他有自动重启吗？

没有重启 我现在暂时解决了，首先是把版本从0.16.2升级到了0.16.3，然后再把以前下过的模型文件删了重新再下了次，那个报错维度错误感觉像是文件出的问题，但是我也没有动过，很奇怪

[qinxuye]

可能还是有问题的，vllm 引擎应该已经死掉了，不过自动杀掉重启的机制没生效，可能还要解决下。

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 5 days since being marked as stale.

[leoterry-ulrica]

可能还是有问题的，vllm 引擎应该已经死掉了，不过自动杀掉重启的机制没生效，可能还要解决下。

v1.1.0版本也复现了这个问题。通过xinference的webui已无法关闭模型，而且直接kill掉模型所占用的进程再launch模型无法启动，最后只能重启xinference服务。

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 5 days since being marked as stale.